diff --git a/src/ra.hrl b/src/ra.hrl
index ceab8819..e6a560b6 100644
--- a/src/ra.hrl
+++ b/src/ra.hrl
@@ -43,7 +43,11 @@
 -type ra_server_id() :: {Name :: atom(), Node :: node()}.
 
 %% Specifies server configuration for a new cluster member.
+%% Both `ra:add_member` and `ra:start_server` must be called with the same value.
 -type ra_new_server() :: #{id := ra_server_id(),
+
+                           %% If set, server will start as non-voter until later promoted by the
+                           %% leader.
                            init_non_voter => ra_nvid()}.
 
 -type ra_peer_status() :: normal |
diff --git a/src/ra_server.erl b/src/ra_server.erl
index 5289ad62..ae2f755c 100644
--- a/src/ra_server.erl
+++ b/src/ra_server.erl
@@ -301,7 +301,7 @@ init(#{id := Id,
             undefined ->
                 InitialMachineState = ra_machine:init(Machine, Name),
                 Clu = make_cluster(Id, InitialNodes,
-                                   #{voter_status => new_voter_status(Config)}),
+                                   #{voter_status => init_voter_status(Config)}),
                 {0, Clu,
                  0, InitialMachineState, {0, 0}};
             {#{index := Idx,
@@ -310,7 +310,7 @@ init(#{id := Id,
                machine_version := MacVersion} = Snapshot, MacSt} ->
                 ClusterNodes = maps:get(cluster_state, Snapshot, ClusterNodes0),
                 Clu = make_cluster(Id, ClusterNodes,
-                                   #{voter_status => new_voter_status(Config)}),
+                                   #{voter_status => init_voter_status(Config)}),
                 %% the snapshot is the last index before the first index
                 %% TODO: should this be Idx + 1?
                 {Idx + 1, Clu, MacVersion, MacSt, {Idx, Term}}
@@ -408,7 +408,7 @@ handle_leader({PeerId, #append_entries_reply{term = Term, success = true,
             Peer = Peer0#{match_index => max(MI, LastIdx),
                           next_index => max(NI, NextIdx)},
             State1 = put_peer(PeerId, Peer, State0),
-            Effects00 = maybe_promote_voter(PeerId, State1, []),
+            Effects00 = maybe_promote_peer(PeerId, State1, []),
             {State2, Effects0} = evaluate_quorum(State1, Effects00),
             {State, Effects1} = process_pending_consistent_queries(State2,
                                                                    Effects0),
@@ -1299,7 +1299,7 @@ handle_receive_snapshot(#install_snapshot_rpc{term = Term,
                                         commit_index => SnapIndex,
                                         last_applied => SnapIndex,
                                         cluster => make_cluster(Id, ClusterIds,
-                                                                #{voter_status => new_voter_status(State0)}),
+                                                                #{voter_status => init_voter_status(State0)}),
                                         machine_state => MacState}),
             %% it was the last snapshot chunk so we can revert back to
             %% follower status
@@ -2368,7 +2368,7 @@ apply_with({Idx, Term, {'$ra_cluster_change', CmdMeta, NewCluster, ReplyType}},
                            [log_id(State0), maps:keys(NewCluster)]),
                     %% we are recovering and should apply the cluster change
                     State0#{cluster => NewCluster,
-                            voter_status => current_voter_status(NewCluster, State0),
+                            voter_status => maybe_promote_self(NewCluster, State0),
                             cluster_change_permitted => true,
                             cluster_index_term => {Idx, Term}};
                 _  ->
@@ -2505,7 +2505,7 @@ append_log_leader({'$ra_join', From,
                    #{id := JoiningNode, voter_status := Voter0},
                    ReplyMode},
                   State = #{cluster := OldCluster}) ->
-    case normalize_voter_status(Voter0, State) of
+    case ensure_promotion_target(Voter0, State) of
         {error, Reason} ->
             {not_appended, Reason, State};
         {ok, Voter} ->
@@ -2525,7 +2525,7 @@ append_log_leader({'$ra_join', From,
 append_log_leader({'$ra_join', From, #{id := JoiningNode} = Spec, ReplyMode},
                   State) ->
     append_log_leader({'$ra_join', From,
-                       #{id => JoiningNode, voter_status => new_voter_status(Spec)},
+                       #{id => JoiningNode, voter_status => init_voter_status(Spec)},
                        ReplyMode}, State);
 append_log_leader({'$ra_join', From, JoiningNode, ReplyMode},
                   State = #{cluster := OldCluster}) ->
@@ -2576,7 +2576,7 @@ pre_append_log_follower({Idx, Term, Cmd} = Entry,
 pre_append_log_follower({Idx, Term, {'$ra_cluster_change', _, Cluster, _}},
                         State) ->
     State#{cluster => Cluster,
-           voter_status => current_voter_status(Cluster, State),
+           voter_status => maybe_promote_self(Cluster, State),
            cluster_index_term => {Idx, Term}};
 pre_append_log_follower(_, State) ->
     State.
@@ -2900,43 +2900,23 @@ already_member(State) ->
 %%% Voter status helpers
 %%% ====================
 
-normalize_voter_status({voter, Reason}, _) ->
+-spec ensure_promotion_target(ra_voter_status(), ra_server_state()) ->
+    {ok, ra_voter_status()} | {error, term()}.
+ensure_promotion_target({voter, Reason}, _) ->
     {ok, {voter, Reason}};
-normalize_voter_status({nonvoter, #{target := _, nvid := _} = Reason}, _) ->
+ensure_promotion_target({nonvoter, #{target := _, nvid := _} = Reason}, _) ->
     {ok, {nonvoter, Reason}};
-normalize_voter_status({nonvoter, #{nvid := _} = Reason}, #{commit_index := CI}) ->
+ensure_promotion_target({nonvoter, #{nvid := _} = Reason}, #{commit_index := CI}) ->
     {ok, {nonvoter, Reason#{target => CI}}};
-normalize_voter_status(_, _) ->
+ensure_promotion_target(_, _) ->
     {error, missing_nvid}.
 
-new_voter_status(#{init_non_voter := NVId}) ->
+-spec init_voter_status(ra_server_config() | ra_new_server()) -> ra_voter_status().
+init_voter_status(#{init_non_voter := NVId}) ->
     {nonvoter, #{nvid => NVId}};
-new_voter_status(_) ->
+init_voter_status(_) ->
     {voter, #{}}.
 
--spec maybe_promote_voter(ra_server_id(), ra_server_state(), effects()) -> effects().
-maybe_promote_voter(PeerID, #{cluster := Cluster} = _State, Effects) ->
-    % Unknown peer handled in the caller.
-    #{PeerID := #{match_index := MI,
-                  voter_status := OldStatus}} = Cluster,
-    case update_voter_status(OldStatus, MI) of
-        OldStatus ->
-            Effects;
-        NewStatus ->
-            [{next_event,
-              {command, {'$ra_join',
-                         #{ts => os:system_time(millisecond)},
-                         #{id => PeerID, voter_status => NewStatus},
-                         noreply}}} |
-             Effects]
-    end.
-
-update_voter_status({nonvoter, #{target := Target} = State}, MI)
-  when MI >= Target ->
-    {voter, State};
-update_voter_status(Permanent, _) ->
-    Permanent.
-
 -spec voter_status(ra_server_state()) -> ra_voter_status().
 voter_status(#{cluster := Cluster} = State) ->
     case maps:get(voter_status, State, undefined) of
@@ -2955,8 +2935,10 @@ voter_status(PeerId, Cluster) ->
             maps:get(voter_status, Peer, {voter, #{}})
     end.
 
-current_voter_status(Cluster, State) ->
-    {_, NReason} = New = voter_status(id(State), Cluster),
+-spec maybe_promote_self(ra_cluster(), ra_server_state()) -> ra_voter_status().
+maybe_promote_self(NewCluster, State) ->
+    %% A bit complicated procedure since nvid is not guaranteed.
+    {_, NReason} = New = voter_status(id(State), NewCluster),
     {_, CReason} = Current = voter_status(State),
     Self = maps:get(nvid, CReason, undefined),
     case maps:get(nvid, NReason, undefined) of
@@ -2964,6 +2946,29 @@ current_voter_status(Cluster, State) ->
         _ -> Current
     end.
 
+-spec maybe_promote_peer(ra_server_id(), ra_server_state(), effects()) -> effects().
+maybe_promote_peer(PeerID, #{cluster := Cluster}, Effects) ->
+    % Unknown peer handled in the caller.
+    #{PeerID := #{match_index := MI,
+                  voter_status := OldStatus}} = Cluster,
+    case update_voter_status(OldStatus, MI) of
+        OldStatus ->
+            Effects;
+        NewStatus ->
+            [{next_event,
+              {command, {'$ra_join',
+                         #{ts => os:system_time(millisecond)},
+                         #{id => PeerID, voter_status => NewStatus},
+                         noreply}}} |
+             Effects]
+    end.
+
+update_voter_status({nonvoter, #{target := Target} = Reason}, MI)
+  when MI >= Target ->
+    {voter, Reason};
+update_voter_status(Permanent, _) ->
+    Permanent.
+
 -spec required_quorum(ra_cluster()) -> pos_integer().
 required_quorum(Cluster) ->
     Voters = count_voters(Cluster),